search

AlexWorldD / NetEmbs

Framework for Representation Learning on Financial Statement Networks
Apache License 2.0
1 stars 1 forks source link

suggestion #5

Closed boersmamarcel closed 5 years ago

boersmamarcel commented 5 years ago

Hi Aleksei,

Currently is see that my data is somewhat noisy in the following form:

0.1A + 0.2A + 0.3A + 0.5B -> C

if another process is:

0.09A + 0.21A + 0.3A + 0.5B -> C

then it is considered as a unique process (this is correct); However, I was wondering what will happen if we slightly simplify the two records to:

0.5 A + 0.5B -> 1C

because currently the algorithm clusters the above two suggestions, which is good, but I'm also curious to see what happens when we simplify the journal entry structure :)

What do you think?

Kind regards,

Marcel

AlexWorldD commented 5 years ago

try prepare_data(..., merge_fa=True, ...) PS - I cannot test it because we don't have in simulated data that situation.